def create_metrics_for_article(article_id, preview=False):
global metricized_articles, terms_processed
# Get article from the DB...
analyzed_article = read_analyzed_article_from_db(article_id)
# Create metrics...
if analyzed_article:
# There are a few different approaches to consider when writing
# metric data.
# (1) Ensure documents are allocated, and then upsert data for
# the change
# (2) Upsert an entire doc each time, where all values are zero
# except one.
# (3) Upsert daily term docs one at a time, then aggregate into
# higher level data later
#
# >> Currently selecting to do the former approach. Results in
# more small reads to the DB, but smaller writes.
# Get needed date values
published = analyzed_article["published"]
yyyy = published.year
mm = published.month
dd = published.day
first_of_month = datetime(yyyy, mm, 1)
days_in_curr_month = date_util.get_days_in_month(yyyy, published.month)
# Iterate over each term in the term histogram
term_histogram = analyzed_article["term_histogram"]
for term in term_histogram:
terms_processed += 1
if not preview:
update_daily_metrics(term, yyyy, mm, dd, first_of_month,
days_in_curr_month, term_histogram[term])
update_monthly_metrics(term, yyyy, mm, term_histogram[term])
# Increase count and update status after each article...
metricized_articles += 1
if preview or metricized_articles % updt_freq == 0:
print " * Articles Metricized: %d..." % metricized_articles
print " Terms: %d Daily Docs %d Monthly Docs %d" % \
(terms_processed, docs_created_daily, docs_created_monthly)
print " Monthly: Read: %s, Create: %s, Write: %s" % \
(mr_time, mc_time, mw_time)
print " Daily: Read: %s, Create: %s, Write: %s" % \
(dr_time, dc_time, dw_time)
else:
print " ERROR: No document with id of '%s' in DB" % article_id
def get_daily_term_data(term, time_start, time_end, granularity):
""" Get 'daily' granularity metric data for the given term and
time range
"""
data = []
avg = 0
total = 0
max_val = 0
data_pts = 0
# Query the DB for all documents containing the
# given term and within the given date range.
db_query = {"term" : term,
"date" : {"$gte" : time_start},
"date" : {"$lte" : time_end}}
result_set = []
result_set_idx = 0;
for result in db_metric_data_daily.find(db_query).sort("date"):
result_set.append(result)
# Iterate over each month in the requested time range
curr_month = time_start
while (curr_month < time_end):
# Typically the end day is the number of days in the month.
# However, if this is the final month in the requested time
# range, the day may be earlier based upon the end date.
days_in_month = date_util.get_days_in_month(
curr_month.year, curr_month.month)
if date_util.is_same_month(curr_month, time_end):
days_in_month = time_end.day
data_pts += days_in_month
# Get the next result from the result set (assume idx is in
# bounds). If it matches the current month, then use its data
# to build the response. Otherwise, we have a gap in the data
# which should be filled with an empty result.
result = None
if result_set_idx < len(result_set):
result = result_set[result_set_idx]
if result and date_util.is_same_month(curr_month, result["date"]):
# Increment result_set_idx, if not
# already at the end of the list
result_set_idx += 1 if result_set_idx < len(result_set) - 1 else 0
for day in range(curr_month.day, days_in_month+1):
val = result["daily"][str(day)]
data.append(val)
total += val
if (val > max_val):
max_val = val
else:
for day in range(curr_month.day, days_in_month + 1):
data.append(0)
curr_month = date_util.get_next_month(curr_month)
# Calculate the final average
if data_pts > 0:
avg = total / data_pts
return data, total, avg, max_val
